What's the difference? Contrasting modular and neural network approaches to understanding developmental variability

Understanding why development differs across individuals is an important challenge for developmental theory. This paper evaluates two approaches to developmental variability observed in domains such as language processing and across populations such as typically developing children, children with developmental disorders, and typical adults. Modular accounts attribute developmental variability to delay, damage, or dysfunction in discrete underlying structures. Neural network approaches attribute developmental variability to emergent effects of graded variations in an interactive, developing system. The authors conclude that neural network approaches offer more formal and parsimonious accounts of the nature and sources of developmental variability.